Sir

Your feature on bioprospecting (Nature 392 535; 1998) conveys a misleading impression of combinatorial libraries: “At present, all these synthetic molecules are polymers”. Although early libraries were mainly based on peptides and nucleotides, the current focus1 is on ‘small molecules’ that are not polymeric and have drug-like physicochemical properties.

I have previously pointed out a fundamental difference between combinatorial chemistry and natural product screening: “At the present time, libraries are constructed by short sequences of synthetic transformations, whereas natural products are the results of much longer and more creative pathways”2. Libraries would approach the structural intricacy of natural products if combinatorial syntheses of ten or more steps were practical. Given the current pace of development, this may well be possible in the future. It is not just a question of increased efficiency, though — chemists string together familiar reactions, while nature is adept at assembling unusual ring skeletons and functional groups.

For example, the chemistry of beta-lactams and enediynes was virtually unknown until their presence was recognized in antibiotics. However, complexity is not always necessary nor desirable. Many medicines are made in a few steps by straightforward reactions, and there is no reason why combinatorial chemistry in its present state cannot lead to many more such drugs.

Anyone engaged in natural product screening will be painfully aware of its bottlenecks. Improvements occur in small increments, unlike the exponential progression of combinatorial chemistry which has led to its enthusiastic reception by the pharmaceutical industry.

The major effect of such comparisons has been to provide a long overdue push to innovate and revolutionize the process of natural product screening, which will ultimately benefit drug discovery as a whole.